Atomization and Flames in LOX=H2-and LOX=CH4-Spray Combustion

Baoe Yang∗

Northwestern Polytechnical University, 710072 Xi’an, People’s Republic of China

Francesco Cuoco†

Avio S.P.A., 00034 Colleferro, Italy

and

Micale Oschwald‡

Institute of Space Propulsion, 74239 Hardthausen, Germany

DOI: 10.2514/1.26538

Hydrogen is a widely used rocket fuel and methane is particularly of interest in Europe as a promising substitute

for H2. Experimental investigation of cryogenic reactive coaxial sprays with oxygen as an oxidizer and hydrogen

and methane as fuels is conducted to prove whether concepts from theLOX=H2 injector design can be transferred

toLOX=CH4 injection. The liquid oxygen has been atomized in shear coaxial atomizers, and the sprays and flames

have been investigated by visualization methods such as shadowgraphy and imaging of the flame emission. LOX

sprays are characterized for both propellants by the intact core lengths and droplet numbers, and the combustion

is analyzed in terms of the flame anchoring mechanism and the flame spreading angle. The results for LOX=H2-

andLOX=CH4-spray combustion are compared, and the influence of the injection conditions of the propellants on

atomization and spray flame is discussed. Significant differences of the sprays and flames have been observed for

the two propellant combinations at similar injection conditions, as defined by the Weber number and the

momentum flux ratio. The flame stabilization process has shown a strong influence on the atomization and flame

characteristics.

Nomenclature

a = thermal diffusivityd, D = diameterJ = gas-to-liquid momentum flux ratio, J � ��u2�g=��u2�lL = intact liquid core lengthOh = Ohnesorge number, Oh� �l=

������������

��ldlp

Re = Renolds number, Re� �ud=�Rv = gas-to-liquid velocity ratio, RV � ug=ulS = flame speedt = LOX post thicknessu = velocityWe = Weber number based on relative velocity,We�

�g�ug � ul�2dl=�� = flame thickness� = viscosity� = density� = surface tension

Subscripts

F = flamef = fuelg = gasL = laminarl = liquido = oxygen

I. Introduction

H YDROGEN is widely used as the fuel of liquid rocket enginesbecause it delivers the highest specific impulse compared to

other fuels, including hydrocarbons. This advantage ofH2, however,is compromised in part by other performance characteristics as, forexample, its cryogenic and low-density properties, costs, anddifficulty of handling. Other noncryogenic propellants like thehypergolic MMH=NTO are toxic where nontoxic propellantsubstitutes are most desirable. For this reason hydrocarbons havecome into focus in Europe as fuels for rocket propulsion, amongthese, CH4 and kerosene are particularly of interest. The mainadvantages of using hydrocarbons are the high propellant density,reduced handling effort, and reduced safety precautions.

Propellant injectors are key components controlling liquid fuelatomization, mixing, combustion, and thus by amajor part efficiencyand stability of combustion in rocket engines. For LOX=H2

combustion chambers, the standard injection element is the shearcoaxial injector as shown in Fig. 1. Liquid oxygen is injected throughthe central tube and the gaseous fuel through the annular slit.

The atomization in coaxial injection is due to a complexinteraction of several forces. Turbulence in the liquid jet aswell as theadaptation of its velocity profile to the new boundary conditionswhen the liquid jet leaves the injector may result in distortions of theliquid jet surface. These surface distortions which are amplified byaerodynamic forces of the high-speed annular gas flow, along withliquid jet destabilization, will result in final jet disintegration.Viscosity and inertia of the liquid have damping effects on thedisintegration dynamics. Because of the complexity of thedisintegration the basic mechanisms leading to atomization areneither completely identified nor well modeled [1]. Therefore thedesign of injectors is mainly based on empirical correlations that usenondimensional parameters characterizing the relative importance ofthe contributing forces. These nondimensional numbers are formedfrom the geometrical characteristics of the injector and the fluid-dynamic properties of the propellants.Most prominent in this contextare the Weber number, the ratio of the aerodynamic force to thesurface tension,

Received 14 July 2006; revision received 31 January 2007; accepted forpublication 4 February 2007. Copyright © 2007 by the authors. Published bythe American Institute of Aeronautics and Astronautics, Inc., withpermission. Copies of this paper may be made for personal or internal use, oncondition that the copier pay the $10.00 per-copy fee to the CopyrightClearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; includethe code 0748-4658/07 $10.00 in correspondence with the CCC.

∗Ph.D. Candidate, College of Astronautics, PO Box 15-11, 710010 Xi’an.†Ph.D. Candidate, Space Propulsion.‡Doctor, Head of Rocket Propulsion, DLR Lampoldshausen. Member

AIAA.

JOURNAL OF PROPULSION AND POWER

Vol. 23, No. 4, July–August 2007

763

We��g�ug � ul�2dl

�

and the momentum flux ratio

J���u2�g��u2�l

TheWeber number was used by Farago and Chigier [2] to classifythe atomization process according to morphology of the dis-integrating liquid jet. They found that the Weber number controlswhether the atomization process is in the Rayleigh breakup regime(We < 25), and it shows a membrane-type breakup (25<We<100) or a fiber-type breakup (100<We < 500). Based on water/airexperiments Engelbert et al. [3] pointed out the influence of the localWeber number on the droplet size instead of theWeber number in theinjector exit plane. These experiments further showed the influenceof the momentum flux ratio on the breakup length. Lasheras et al. [4]derived for the dependence of the breakup length L on J:

L

dl� 6

���

Jp (1)

Recent research on coaxial jet breakup lengths conducted byDavis [5] and Woodward et al. [6] showed considerably longerbreakup lengths than predicted by the above correlation of Lasheraset al. The correlation derived by Davis from cold-flow experimentsfor two-phase coaxial jets was

L

dl� 25

J0:2(2)

For single-phase coaxial jets in supercritical regime:

L

dl� 12

J0:5

Woodward et al. suggested a ln �J� scaling for correlating the LOXdense core length from their limited LOX=H2 hot fire data atsupercritical conditions.

There are other nondimensional groups related to liquid jetatomization, for instance, the velocity ratio, the Ohnesorge number,and the liquid Reynolds number. The velocity ratio RV � ug=ul hadbeen found in experiments to be a nondimensional groupcharacterizing LOX=H2-spray flames in respect to combustionstability [7] or combustion efficieny [8]. Drop breakup regimes incold flows [9,10] have been classified by the Ohnesorge numberOh� �l=

������������

��ldlp

. As a ratio of inertial and viscosity, liquid jetReynolds number may also play a role in jet disintegration as Eroglu

and Chigier [11] observed for airblast coaxial atomizers at lowWeber number (We < 300).

The relative importance of the forces involved in the atomizationprocess strongly depends on the injection conditions, the physicalproperties of the propellants, and the localflowfield in the spray. Thisis the reason why even in nonreacting sprays correlations derivedfrom experimental results only show a rough agreement, forexample, there are inconsistencies on which property increases ordecreases droplet size [8]. An extrapolation of correlations obtainedin cold-flow tests with water to cryogenic conditions with the lowlevel of surface tension and viscosity of liquid oxygen is thereforehighly unreliable. This is due to the fact that theReynolds numbers ofliquid oxygen are generally 1 order of magnitude belowrepresentative conditions (Re� 105–106) when using substitutefluids such as water.

For cold-flow coaxial injection, there are numerous experimentaland theoretical investigations [1–5,10–16]. The major question,however, is whether and how results from cold-flow tests can betransferred to hot fire conditions. Unfortunately there are not verymuch data with systematic parameter variation for reactive spraysavailable. In Fig. 2, an image of the emission of the OH radical in theflame of coaxial LOX=H2 spray is shown. It is clearly seen that theflame is stabilized in thewake of the rim of the central LOX tube, alsoknown as LOX post. As a consequence, the LOX jet is separatedfrom the annular H2 high-speed gas flow by the reaction front and aturbulent mixing layer of hot reaction products and reactants.Combustion in hot fire tests has therefore several consequences forthe atomization process.

For an attached flame, the liquid oxygen jet is not directly exposedto the aerodynamic forces of the annular gaseous fuel flow, but theseforces have to be transmitted by the turbulent mixing layer to theliquid surface. As a consequence the spray formation may happenunder a different atomization regime as compared to cold-flowconditions. Visualization of LOX sprays at identical injectionconditions in cold flow and hot fire tests for instance clearly showsthe faster evaporation of the small LOX droplets in the reactive flow.

From this discussion, the limitations of cold-flow tests inpredicting hot fire atomization become evident.When discussing therole of the fuel properties on the atomization process it is not enoughto focus only on the physical properties of the fuel like its density,injection velocity, etc., but the kinetic properties of the reactionpartners, and the turbulent transport properties have to be taken intoaccount as well.

Flame anchoring at the LOX post as shown in Fig. 2a has beenconsistently observed for LOX=H2 injection [17–20]. However witha different fuel type such as CH4, a lifted flame anchoring in theturbulent mixing layer of evaporated LOX and gaseous fuel is also apossible flame anchoring mechanism (see Fig. 2b). In this case theinteraction of combustion with the spray formation process startsdownstreamof theflame anchoring position. Juniper andCandel [21]derived from numerical investigations that the nondimensionalquantity � t=�F, the ratio of LOX-post thickness t and the flamethickness �F, is a control parameter for the flame stabilizationbehavior.

There is a huge database on LOX=H2 combustion in Europe. It isworthwhile then to compare LOX=CH4- to LOX=H2-spray

Fig. 1 Sketch of a coaxial injector.

LOX

H2

H2

LOXCH4

CH4

a) LOX/H2-flame anchored at the LOX-post b) Lifted flame anchoring in the O2/CH4 shear layer

Fig. 2 Different flame anchoring mechanisms. Shown is the imaging of OH emission a) observed in a model combustor at the test bench P8 of DLR

(German Aerospace Center), and b) this work.

764 YANG, CUOCO, AND OSCHWALD

combustion to prove whether concepts from LOX=H2-injectordesign can be transferred to LOX=CH4 injection. Comparing thethermophysical properties it can be assumed that there may belimitations. Some properties of the propellants at the test conditionsof this paper are listed in Table 1 where differences can be identified.At the injection conditions the density of methane is more than2 times the density of H2 and the specific heat only about 0.2 timesthat of H2 at typical injection conditions. The combination of theseproperties shows the impact on the flame through thermal diffusivity.The flame thickness �F is related to the laminar flame speed SL andthe thermal diffusivity a by the following equation: �F � 2a=SL[26]. The laminar flame speed for CH4=O2 is about a factor of 2.5below the value for H2=O2, which may be of great importance forflame propagation and stabilization during the ignition transient andflame anchoring phenomenology at stationary conditions. And theremarkable difference in ignitability in the fuel rich limit has to bepointed out.

The experimental work presented in this paper is dedicated toshow the influence of the fuel properties and its kinetics onatomization and combustion by comparing LOX=CH4- andLOX=H2-spray combustion phenomenology.

II. Test Setup

A. Test Bench and Combustor

The experimental investigation had been performed on theM3 testfacility at DLR-Lampoldshausen, Germany. The test bench isequipped with a single-injector combustion chamber (see Fig. 3)with 140 � 40 mm2 quartz windows for optical access forvisualizing the spray and flame development. Combustion can beinitiated in two ways: a torch igniter or Nd:YAG-laser inducedplasma. The latter method produces precise ignition energy, timing,and positioning. Propellants mass flow rates were chosen to achieve0.15 MPa combustion chamber pressure once steady-stateconditions are reached. The duration for each test was 2 s becausethe optical windows were uncooled. The Weber and J-numbercombinations were varied in the investigation from 500–16,000 and

from 0.2 to 2.0, respectively, via changing the LOX post andfaceplate geometry of the coaxial injector. The main parameters ofinjector geometries and test conditions were summarized in Table 2.To check the experimental reproducibility, at least two tests wereconducted for one experimental test case.

The experimental conditions are representative of real engines inrespect to the temperature of the cryogenic propellants. Regardingthe J and Weber numbers, the experiments reflect the lowerboundaries of representative conditions for LOX=H2-rocketcombustors. The chamber pressure, limited to 0.15 MPa in thesetests by the experimental setup, mainly influences the surface tensionof liquid oxygen and the coaxial gas density. In spite of the chamberpressure limitation, the results presented may help to understand theinteraction between atomization and flame characteristics and toprove whether concepts from LOX=H2-injector design can betransferred to LOX=CH4 injection.

B. Flow and Flame Visualization Techniques

A high-resolution shadowgraph setup was used for obtaining theliquid oxygen spray information. Shadowgraphs have been recordedusing a Kodak Flowmaster 2k camera with a high spatial resolutionof 0:055 mm=pixel and 4 frames=s acquisition rate. The flowfieldwas “frozen“ by means of a back light from a nanolite with a 18 nsflash duration.

Flame development was visualized with an intensified high-speedcharge-coupled device (CCD) camera with a 9 kHz acquisition rateand a 256 � 128 pixel resolution. The camera was fitted with a UVlens and a narrowband filter (300–310 nm) to record the OH-radicalemission during the combustion process.

III. Data Reduction

A. Spray Information from Shadowgraph Images

1. Liquid Intact Core Length

Liquid intact core length, which is also called “breakup length” or“decay length,” refers to the distance from an injector exit surface tothe point a continuous jet breaks up. This is the most straightforwardspray information that can be extracted quantitatively fromshadowgraph images. Images are processed as shown in Fig. 4 byconverting an intensity image into a binary image and labeling thedisconnected components as different objects. The most upstreamobject is taken as the liquid intact core.

The dependence of intact core length determination on thethreshold value has been tested on three typical images as shown inFig. 5. If the threshold value is chosen to be too small, bright liquid isidentified as background, and hence the evaluated core length wouldbe too small. If the threshold is too high, then the evaluated intact corelength will be too large. The threshold value is the most criticalparameter in the process of intact core length determination. Thereexists a threshold range in which the discrimination between liquidand background is not sensitive to the threshold. In Fig. 5 the intactcore lengths determined from one image vary less than 3% for a

Table 1 Thermophysical properties of propellants at injector exit

conditions of this paper: P� 0:15 MPa, T � 80 K for LOX andH2,

ambient temperature for CH4 [22,23]

O2 CH4 H2

Critical temperature, K 154.6 190.5 32.9Critical pressure, MPa 5.04 4.60 1.28Reduced pressure, P=Pcrit 0.030 0.033 0.12Reduced temperature, T=Tcrit 0.52 1.51 2.43Density, kg=m3 —— 1.01 0.456Viscosity, �Pa � s —— 10.86 3.57Specific heat, J=kg � K —— 2213 10,766Thermal conductivity, W=m � K —— 0.0329 0.0566Thermal diffusivity, 10�5 m2=s —— 1.47 1.15Laminar flame velocity @ ambient[24], m=s

—— 3.93 10.7

Ignitability limits [25], Vol % —— 5.1–61 4–94

Fig. 3 The M3 combustor.

Table 2 Injector geometries and test conditions

LOX=H2 injector LOX=CH4

injector

LOX-post inner diameter do,mm

1.2 1.6 1.4, 1.6

LOX-post thickness t, mm 0.4 0.4

Annular slit diameter df ,mm

4.8, 5.2,5.7

4.9, 5.7,6.9

5.0, 5.7,6.9

LOX inlet temperature, K 80 80

Fuel inlet temperature, K 80 Ambienttemperature

LOX mass flow rate, g=s 8–20 11–24

Propellant mixture ratio 5.5 3.4

Combustion pressure, MPa 0.15 0.15

YANG, CUOCO, AND OSCHWALD 765

variation of the threshold level between 10 and 35. The specificthreshold for each test condition has been chosen individuallydepending on the image characteristics and is typically in the range of25–30.

During stationary flow conditions the length of the intact corevaries due to the intrinsic instability of the liquid jet atomizationprocess. The variation of the core lengths evaluated from differentimages in one test is therefore much higher than the error due to thechosen threshold level. For instance, the standard deviation of theintact core lengths measured in one test at conditions of Fig. 4 is20.3% of the mean but the error from thresholds is less than 3% asmentioned above.

2. Liquid Droplet Number

The objective of liquid jet atomization is to break up a jet intodroplets as fine as possible to maximize the liquid surface area.Although we cannot get droplet sizes from shadowgraph images,numbers of droplets discernible in the shadowgraph images may becounted as a quantitative way to characterize the atomizationprocess.

The raw images have to be converted into binary images to labeldroplets, including the jet itself and other large liquid structures, asdifferent objects. The results are more sensitive to image processingas compared to the determination of the liquid core length. Dropletshave to be discriminated against the background in a much largerimage area with a variation of the background level as compared to

the liquid intact core. Contrast enhancement has been done on imagesubareas which then have been converted into binary images (see thetop images of Fig. 6). The subareas and the threshold level have beentuned by the operator’s judgment so that the binary images reflect themain information observed in the original image. Droplet numbersobtained by this process appeared to vary by about 12% in theselected threshold range. The major contribution to this errororiginates from the image noise.

Because of the spatial resolution limitation (55 �m) of theshadowgraph images, only drops at least larger than 1–2 pixel sizecan be identified. This introduces systematic errors in thedetermination of droplet numbers. At the same time, the discernibledrops are very different in their sizes and this information ofatomization cannot be reflected by the droplet numbers. With theselimitations inmind a qualitative comparison of the results obtained indifferent tests is done. Although the droplet-number information isonly a rough statistic of the atomization process it appeared to be veryeffective in analyzing the Weber and J effects on atomization asshown in the next section.

B. Flame Information from OH Images

OH-emission images have been processed to obtain the flamespreading angle. A flame angle is estimated from the average of OHimages in one test run. The flame boundary is determined byseparating the flame from the background by using an adequatethreshold to remove image noise such as signals due to reflectionfrom the combustor walls (see Fig. 7). Two lines are fitted to theboundary of the conical flame surrounding the LOX spray. Half ofthe angle between these two lines represents the flame spreadingangle. Based on the sensitivity of the spreading angle on varying thethreshold in the grayscale range of 45–55, we estimated an accuracyof about 4% of the angle.

IV. Results and Analysis

A. Sprays

1. Qualitative Comparison of Sprays in LOX=H2 and LOX=CH4 Hot

Fire Tests

As shown in Figs. 8–10, liquid oxygen sprays behave verydifferently in LOX=H2 and LOX=CH4 hot fire tests. If coaxial sprayatomization is mainly controlled by the dimensionless numbers

a) Intensity image (LOX/H2, We=9844, J=1.28) b) Binary image

Fig. 4 Processing of shadowgraph images for liquid intact core length.

0 10 20 30 40 50 6020

25

30

35

40

45

50

55

60

L/d

o at

Dif

fere

nt T

hres

hold

s

Threshold Grayscale (Total Range [0,255])

Image1 Image2 Image3

Fig. 5 Effect of thresholds ondetermination of liquid intact core length.

(LOX/H2, We =9844, J=1.28)

(LOX/CH4, We =5444, J=0.91)

a) Processed intensity image b) Binary image

Fig. 6 Processing of shadowgraph images for droplet numbers.

766 YANG, CUOCO, AND OSCHWALD

Weber and J as supposed before, the atomization scenes should besimilar at similar Weber and J conditions. What is found in theexperiments is contrary to this, with the jet surface being muchwavier, and more ligaments and droplets form around the wavyliquid core in LOX=CH4 compared to LOX=H2 tests. Thedimensions of blobs and ligaments are smaller and droplets can befound at larger radial distances from the jet for LOX=CH4. Larger

amplitude of the lateral pulsation of the spray and earlier jet surfaceinstability development results in shorter liquid intact core length inCH4 sprays. These phenomena become more obvious withincreasing Weber and J numbers and the spray shows dramaticdifferences especially at high Weber and J numbers as shown inFig. 10. Test conditions for the images in Figs. 8–10 are listed inTable 3.

a) Average image of the flame b) Flame angle calculation

Fig. 7 Processing of OH images for flame angles.

a) We=2208, J=0.3 c) We=1843, J=0.22

b) We=2462, J=0.73 d) We=2498, J=0.95

Fig. 8 J effect at the Weber number around 2000 for a), b) LOX=H2 tests, and c), d) LOX=CH4 tests.

a) We=5142, J=0.68 c) We=5444, J=0.91

b) We=5014, J=1.46 d) We=8023, J=1.67

Fig. 9 J effect at the Weber number around 5000 for a), b) LOX=H2 tests, and c), d) LOX=CH4 tests.

a) We=9844, J=1.28 b) We=9885, J=1.62

Fig. 10 Comparison at the Weber number around 10,000 for a) LOX=H2 test, and b) LOX=CH4 test.

YANG, CUOCO, AND OSCHWALD 767

Although the comparison of LOX-spray behavior inLOX=H2 andLOX=CH4 hot fire tests shows that with the presence of combustion,atomization becomes more complex and there must be othercontrolling parameters besides Weber and J, the two numbers stillshow their clear effects on atomization. In Figs. 8 and 9 the effect ofthe J number at similar Weber number can be seen for a Webernumber around 2000 in Fig. 8 and near to 5000 in Fig. 9. Higher gas/liquid momentum flux ratio J has the tendency to promote jetinstability and to result in an earlier onset of jet disintegration. ThisJ effect keeps working at different Weber conditions both in H2 andCH4 tests.

If Figs. 8b and 8d are compared with Figs. 9a and 9c, and Figs. 9band 9d with Figs. 10a and 10b, respectively, the Weber effect can beseen at J� 0:7–0:9 and J� 1:3–1:6. As a ratio of aerodynamic andsurface tension forces, the Weber number represents the relativedisturbance of the gaseous coflow to the jet surface. With increasingWeber number, ligament and droplet dimensions become smallerand secondary atomization is augmented, especially shown in theimages of LOX=CH4 tests. The effect results in finer droplets atshorter downstream distance from the injector exit. Droplets are alsotransported to larger radial distances from the spray axis at higherWeber number, an effect particularly obvious in LOX=CH4 tests.The very dark upstream region in Fig. 8d is due towater condensationon the optical windows, which has been processed to enhance thecontrast of the region.

After this qualitative description of sprays in LOX=H2 andLOX=CH4 hot fire tests, a quantitative analysis is given to supportthe arguments above.

2. LOX Intact Core Length

Processing the shadowgraphs as described in Sec. III, the intactcore lengths of the liquid oxygen sprays in LOX=H2 and LOX=CH4

hot fire tests at different injection conditions are determined andshown in Figs. 11a and 11b respectively. The x axis represents J

number and the color bar shows a rough scale of the Weber number.The uncertainty of the values is about 20% for the J number andabout 10% for the Weber number.

The evaluated LOX intact core length presents a great variationeven at very similar conditions. This is the reasonwhyonly a linearfitis given just to show the tendency of the effect of J. The standarddeviation of the core length from the mean at a condition is alsoshown in Fig. 11. Considering the development of the jet instabilityas the reason of jet disintegration, it is reasonable to assume that theposition of the primary disintegration point is pulsating with certainamplitude. What have been recorded in the images are differentphases of jet pulsation and result in the variation of the intact corelength during one test.

It is clearly seen that the LOX intact core length in LOX=H2 testsare much longer than in LOX=CH4 tests at similar Weber and Jconditions (more than 30% longer at J < 0:5 and 20% longer atJ� 1:5). The results of Porcheron et al. [15] show the amplifyingeffect of gas/liquid density ratio on primary jet disintegration and acorrelation is given as liquid core lengthL=Do / ��g=�l��0:38 basedon experimental results. This result gives an indication that the muchhigher density ofCH4 as compared toH2 (see Tables 1 and 3) may bethe reason of shorter intact core length for CH4 tests even wheninjection conditions in terms of Weber and J numbers are similar.The higher gaseous Reynolds numbers in LOX=H2 tests seem to notpromote the primary breakup of the LOX jet.

The mean intact core length is decreasing with increasing Jnumbers. TheWeber number does not show a similar obvious effecton the intact core length as J. At relative low Weber number(We <�5000), the LOX core length in H2 tests shows a slightlyinverse relationshipwithWeber, but at highWeber the effect tends todiminish in both tests. The J and Weber effects on intact core lengthhave been found to be similar in H2 and CH4 tests.

The intact core length calculated from Eq. (1) of Lasheras et al. [4]and Eq. (2) of Davis [5] is comparedwith the core lengthmeasured inthis work in Fig. 11. Results predicted by Lasheras are almost 1 order

Table 3 The detailed test conditions for images in Figs. 8–10

Test Dl, mm Dg, mm We J ug, m=s ul, m=s Reg Rel �g=�l

LOX=H2

Figure 8a 1.6 6.9 2208 0.30 226 8.2 138,200 56,200

3:9e-4Figure 8b

1.6 5.72461 0.73 235 5.5 105,600 37,500

Figure 9a 5142 0.68 340 8.2 152,800 56,200

Figure 9b1.6 4.9

5014 1.46 333 5.5 113,400 37,500Figure 10a 9844 1.28 467 8.2 159,000 56,200

LOX=CH4

Figure 8c 1.4 6.9 1843 0.22 159 9.6 64,100 57,600

8:0e-4Figure 8d

1.6 5.72498 0.95 167 4.9 47,400 33,300

Figure 9c 5444 0.91 247 7.4 70,100 50,500

Figure 9d1.6 5.0

8023 1.67 228 4.9 51,000 33,300Figure 10b 9885 1.62 330 7.4 73,800 50,500

0.0 0.5 1.0 1.50

10

20

30

40

50

60

70

LO

X in

tact

cor

e le

ngth

L/d

o

Momentum flux ratio J

linear fit Davis[5]: 2-phase Lasheras et al.[4]

0

2000

4000

6000

8000

10000

12000

We number

0.0 0.5 1.0 1.5 2.00

10

20

30

40

50

60

70

LO

X in

tact

cor

e le

ngth

L/d

o

Momentum flux ratio J

linear fit Davis[5]: 2-phase Lasheral et al.[4]

0

2000

4000

6000

8000

10000

12000

We number

a) LOX/H2 tests b) LOX/CH4 Tests

Fig. 11 J and Weber effects on liquid intact core length with standard deviations.

768 YANG, CUOCO, AND OSCHWALD

shorter than the core lengths of Davis and this work. Davis’scorrelation is in good agreementwith the LOXcore length in theCH4

hot tests.

3. Droplet Numbers

The evaluated droplet numbers of liquid oxygen sprays inLOX=H2 and LOX=CH4 hot fire tests are shown in Figs. 12a and12b, respectively.

TheWeber number is chosen as the x axis in Fig. 12; this is due to astrong relationship between the droplet number and Weber numberwith an alteration of the J having a small effect. The correlation ofdroplet number data with the Weber number is better than that ofintact core length with J in Fig. 11. In the graphs it is clearly shownthat more droplets and ligaments form and peel off a jet at higherWeber number both in theH2 andCH4 tests. According to Engelbertet al. [3], droplet formation is mainly characterized by the localWeber number. The result implies higher local Weber number athigher injection Weber due to the augment of bulk velocity in thechamber with increasing injection Weber number.

Considering liquid intact core length and droplet numbers asmanifestations of primary breakup and secondary breakup,respectively, the effects of J andWeber numbers could be explainedfrom the physical meanings of the two parameters. A high gas/liquidmomentum flux ratio J at the injector exit means relative lower liquidjet inertia, which implies reduced tendency of the jet to resistchanges. A jet with lower inertia is easier to disturb and shows anearlier onset of primary jet disintegration. The effect of the initial jetinertia almost fades away after the primary jet disruption and that is

the reason why J has little effect on droplet numbers. The localWeber number controls secondary breakup and results in more andfiner droplets. The quality of secondary breakup does not show aclose relationship with the primary breakup length.

There is a notable difference of droplet numbers in theH2 andCH4

tests. The reason for lower droplet numbers inH2 testsmay be relatedto 1) the higher chemical reaction rate forLOX=H2 resulting in moreconsumption of gaseous oxygen and thus the partial pressuregradient of gaseous O2 is increased. This should enhance theevaporation rate of liquid oxygen. 2) The higher rate of chemical heatrelease promotes the evaporation of oxygen droplets.

The clear effects of Weber and J numbers on LOX sprays and thegreat differences betweenLOX=H2 and LOX=CH4 sprays in hot testimplies the significant influence of combustion properties onatomization behavior.

B. Flame

1. Flame Anchoring

As described in Sec. III, the OH emission within the reaction zonewas recorded to obtain quantitative information on the flamepatterns. Comparing LOX=H2 and LOX=CH4 flames it is found thatLOX=H2 flames are always attached to the injector rim, whereasLOX=CH4 flames are easily lifted off under the conditions tested inthis work. The distance of the flame base position to the injector faceplate can be 8 times the LOX-post diameter for the case in Fig. 13d.

Flame stabilization in the turbulent two-phase flow depends onmany factors. The primary reason for the liftoff potential of CH4

flame is the much slower kinetics compared with H2 (see Table 1).

0 2000 4000 6000 8000 100000

50

100

150

200

250

300

Liq

uid

drop

let n

umbe

r

We number

J number

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0 5000 10000 150000

200

400

600

800

1000

1200

1400

Liq

uid

drop

let n

umbe

r

We number

J number

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

a) LOX/H2 tests b) LOX/CH4 tests

Fig. 12 Weber and J effects on LOX-spray droplet numbers.

a) We=2192, J=0.47 c) We=2335, J=0.60

b) We=7007 J=0.65 d) We=7936, J=0.56 Fig. 13 Weber effect on LOX=CH4-spray flame for a), b) LOX=H2 tests, and c), d) LOX=CH4 tests.

YANG, CUOCO, AND OSCHWALD 769

Figure 13 shows as an example the different flame patterns for thetwo propellant pairs under similar Weber and J combinations.

Lifted flames give their fingerprint also in the shadowgraphimages. For the lifted flames a distinct change in the spray pattern hasbeen observed at the flame anchoring position marked by the arrowsin Fig. 14. Upstream of the anchoring position the spray is ratherconfined in a small angle and few isolated liquid fibers and dropletsare visible. At the anchoring position a sudden increase of thespreading angle of the spray is observed. The sudden expansion ofthe spray may be due to the rapid reactive heat release starting fromthe anchoring point. This clearly demonstrates the influence of theflame anchoring mechanism on the spray pattern.

No significant correlation of the liftoff distance with thenondimensional numbers Weber and J has been found. Althoughone would expect that the liftoff distance is increasing withincreasingCH4 velocity and thuswith increasingWeber number, thisseems not to be the case in our results.

2. Flame Angle

The dependence of the flame angles for LOX=CH4 and LOX=H2

sprayflames on theWeber number is shown in Fig. 15.A comparisonof the results for both propellants shows increasing Weber-numberresults in an increasing flame spreading angle. The methane datashown in the graph can be separated into two groups: one for liftedflames and the other for attached flames. Lifted flames have onlybeen observed for methane and flame angles of lifted are muchbroader than for attached LOX=CH4 flames. For attached flamessimilar results are observed for LOX=H2 and CH4.

TheWeber dependence of the flame angle is in agreement with theobserved Weber dependence of droplet numbers and the radialdispersion of droplets. Unlike for Weber number, no significantcorrelation of the flame angle with either the momentum flux ratio orthe velocity ratio has been found. It is worth noticing that the flameangles in single-injector experiments may be broader than flame

angles from multi-injectors tests since there is no confinement byother streamlines from neighboring injectors.

V. Summary and Conclusions

By comparing LOX=H2 and LOX=CH4 spray combustionphenomenology, significant differences of the sprays and flameshave been observed for similar injection conditions as defined byWeber number and momentum flux ratio.

Compared to sprays of LOX=H2 for hot fire tests, LOX=CH4

sprays present shorter liquid oxygen core length, more discernibledroplets, and larger spray dispersion at similar injectionWeber and Jnumbers.

The effect of J andWeber numbers on atomization in hot fire testsis similar to the findings under cold-flow conditions: the J numbermainly governs the primary breakup of the liquid jet and thusdetermines the liquid core length, and the Weber number mainlyinfluences secondary atomization and flame angles.

Anchored LOX=CH4 flames show similar flame angles asanchored LOX=H2 flames. Lifted flames are only observed forLOX=CH4 and these flames show larger flame angles as the attachedflames. This together with the data from the spray visualizationsproves the strong coupling between flame stabilization process,flame characteristics, and atomization.

From the results obtained until now, it is obvious thatnondimensional numbers characterizing the fluid-dynamic inter-action of the two fluids at the injector exit are not sufficient to scalecoaxial injector performance in hot fire tests from one fuel to another.At identical injection conditions in terms of Weber number andmomentum flux ratio, especially the flame stabilization mechanismsmay be different for different fuels. Scaling of injector designs fordifferent types of fuel has therefore to take into account kinetic andtransport properties associated with combustion.

Acknowledgments

The support of Avio S.P.A. to F. Cuoco and of the ChinaScholarship Council to B. Yang is gratefully acknowledged.

References

[1] Ledoux, M., Caré, I., Micci, M., Glogowski, M., Vingert, L., andGicquel, P., “Atomization of Coaxial Injectors,”Proceedings of the 2ndInternational Symposium on Liquid Rocket Propulsion, ONERA,Châtillon, France, 1995, pp. 2-1–2-19.

[2] Farago, Z., and Chigier, N., “Morphological Classification ofDisintegration of Round Liquid Jets,” Atomization and Sprays, Vol. 2,No. 2, 1992, pp. 137–153.

[3] Engelbert, C., Hardalupas, Y., and Whitelaw, J. H., “BreakupPhenomena in Coaxial Airblast Atomizer,” Proceedings of the RoyalSociety of London A, Vol. 451, No. 1941, 1995, pp. 189–229.

[4] Lasheras, J. C., Villermaux, E., and Hopfinger, E. J., “Break-Up andAtomization of a Round Water Jet by a High-Speed Annular Air Jet,”Journal of Fluid Mechanics, Vol. 357, Feb. 1998, pp. 351–379.

[5] Davis, D, W., and Chehroudi, B., “Measurements in an AcousticallyDriven Coaxial Jet Under Sub-, Near-, and Supercritical Conditions,”Journal of Propulsion and Power, Vol. 23, No. 2, 2007, pp. 364–374.

Flame anchoring position

Flame anchoring position

a) We=7936, J=0.56 b) We=9885, J=1.65 Fig. 14 Detail of the spray pattern near the flame liftoff position of LOX=CH4-spray flames.

0 5000 10000 15000 200000

10

20

30

40

50

60 H2

CH4 (anchored flame)

CH4 (lifted flame)

flam

e sp

read

ing

angl

e

WeFig. 15 Flame spreading angle vs Weber number for LOX=CH4 and

LOX=H2 tests.

770 YANG, CUOCO, AND OSCHWALD

[6] Woodward, R. D., Pal, S., Farhangi, S., and Santoro, R. J., “LOX=GH2Shear Coaxial Injector Atomization Studies at Large Momentum FluxRatios,” AIAA Paper 2006-5203, 2006.

[7] Wanhainen, J. P., Parish,H.C., andConrad, E.W., “Effect of PropellantInjection Velocity on Screech in 20000 Pound Hydrogen-OxygenRocket Engine,” NASA TN D-3373, 1966.

[8] Smith, J. J., Schneider, G., Suslov, D., Oschwald, M., and Haidn, O.,“Steady-State High Pressure LOx=H2 Rocket Engine Combustion,”Aerospace Science and Technology, Vol. 11, No. 1, 2007, pp. 39–47.

[9] Levebre, A. H., Atomization and Sprays, Taylor and Francis, London,1989.

[10] Chigier, N., and Reitz, R. D., “Regines of Jet Breakup and BreakupMechanisms (Physical Aspects),” Recent Advances in Spray

Combustion: Spray Atomization and Drop Burning Phenomena,edited by K. Kuo, Vol. 166, Progress in Astronautics and Aeronautics,AIAA, Washington, D.C., 1995, pp. 109–135.

[11] Eroglu, H., and Chigier, N., “Wave Characteristics of Liquid Jets fromAirblast Coaxial Atomizers,” Atomization and Sprays, Vol. 2, No. 3,1992, pp. 295–318.

[12] Rahman, S. A., and Santoro, R. J., “A Review of Coaxial Gas/LiquidSpray Experiments and Correlations,” AIAA Paper 94-2772, 1994.

[13] Villermaux, E., “Mixing and Spray Formation in Coaxial Jets,” Journalof Propulsion and Power, Vol. 14, No. 5, 1998, pp. 807–817.

[14] Hardalupas, Y., and Whitelaw, J. H., “Characteristics of SpraysProduced by Coaxial Airblast Atomizers,” Journal of Propulsion and

Power, Vol. 10, No. 4, 1994, pp. 453–460.[15] Porcheron, E., Carreau, J. L., Prevost, L., Le Visage, D., and Roger, F.,

“Effect of Injection Gas Density on Coaxial Liquid Jet Atomization,”Atomization and Sprays, Vol. 12, Nos. 1–3, 2002, pp. 209–227.

[16] Vingert, L., Gicquel, P., Lourme,D., andMénoret, L., “Coaxial InjectorAtomization,” Liquid Rocket Combustion Instability, edited by V.Yang, Vol. 169, Progress in Astronautics and Aeronautics, AIAA,Washington, D.C., 1995, pp. 145–189.

[17] Mayer, W., Ivancic, B., Schik, A., and Hornung, U., “PropellantAtomization in LOX=GH2 Rocket Combustors,” 34nd AIAA/ASME/

SAE/ASEE Jet PropulsionConference andExhibit, AIAA,Reston, VA,

July 13–15, 1998; also AIAA Paper 1998–3685.[18] Mayer, W., and Smith, J., “Fundamentals of Supercritical Mixing and

Combustion of Cryogenic Propellants,” Liquid Rocket Thrust

Chambers: Aspects of Modeling, Analysis and Design, edited by V.Yang, Vol. 200, Progress in Aeronautics and Astronautics, AIAA,Reston, VA, 2004, pp. 339–367.

[19] Herding, G., Snyder, R., Scouflaire, P., Rolon, C., and Candel, S.,“Flame Stabilization in Cryogenic Propellant Combustion,” Proceed-ings of the 26th Symposium (International) on Combustion, TheCombustion Institute, Pittsburg, PA, 1996, pp. 2041–2047.

[20] Candel, S., Herding, G., Synder, R., Scouflaire, P., Rolon, C., Vingert,L., Habiballah, M., Grisch, F., Péalat, M., Bouchardy, P., Stepowski,D., Cessou, A. and Colin, P., “Experimental Investigation of ShearCoaxial Cryogenic Jet Flames,” Journal of Propulsion and Power,Vol. 14, No. 5, 1998, pp. 826–834.

[21] Juniper, M., and Candel, S., “Edge Diffusion Flame StabilizationBehind a Step over a Liquid Reactant,” Journal of Propulsion and

Power, Vol. 19, No. 3, 2003, pp. 332–341.[22] Younglove, B. A., “Thermo-Physical Properties of Fluids. 1,” Journal

of Physical and Chemical Reference Data, Vol. 11, Suppl. 1, 1982,pp. 97–161.

[23] Younglove, B. A., and Ely, J. F., “Thermo-Physical Properties ofFluids. 2,” Journal of Physical and Chemical Reference Data, Vol. 16,No. 4, 1987, pp. 577–798.

[24] Fiock, E. F., High Speed Aerodynamics and Jet Propulsion, Vol. 9:

Physical Measurements in Gas Dynamics and Combustion, Sec. K:

Measurement of Burning Velocity, Princeton Univ. Press, Princeton,NJ, 1953, pp. 409–438.

[25] Coward, H. F., and Jones, G. W., Limits of Flammability of Gases andVapors, Bulletin 503, Bureau of Mines, 1952.

[26] Turns, S. R., An Introduction to Combustion: Concepts and

Applications, 2nd ed., McGraw–Hill, New York, 2000.

D. TalleyAssociate Editor

YANG, CUOCO, AND OSCHWALD 771